Can AI expand 5 personality scores into a fuller profile?
Researchers asked a simple question: given just your Big Five personality answers, can an AI “role‑play” how you’d score on other psychology questionnaires?
Using responses from 816 people, large language models were prompted to generate answers on nine additional psychological scales. The patterns in the AI’s generated data closely matched real human patterns across scales (R^2 > 0.89). This was achieved zero‑shot, beat semantic similarity baselines, and approached models trained directly on the dataset.
How did the AI do it? Analysis suggests a two‑step process: (1) it turns raw Big Five numbers into a short, natural‑language personality summary (a kind of compressed “gist”), then (2) reasons from that summary to produce answers on new scales. The models spotted the same key factors as trained algorithms, though they didn’t finely weight individual items. Notably, these summaries added predictive value beyond the original scores, hinting they capture emergent, second‑order patterns in how traits interact.
Why it matters: this offers a new tool for psychological simulation and a window into how LLMs abstract and reason. Caveat: it’s a research study—not a clinical tool—and raises privacy and consent considerations.
Paper: http://arxiv.org/abs/2511.03235v1
Paper: http://arxiv.org/abs/2511.03235v1
Register: https://www.AiFeta.com
AI psychology LLM personality BigFive research ethics
